ARTICLE

Tectonically-triggered sediment and carbon exportto the Hadal zoneRui Bao1,2,7, Michael Strasser 1,3,4, Ann P. McNichol2, Negar Haghipour1, Cameron McIntyre 1,5,6,

Gerold Wefer4 & Timothy I. Eglinton1

Sediments in deep ocean trenches may contain crucial information on past earthquake his-

tory and constitute important sites of carbon burial. Here we present 14C data on bulk organic

carbon (OC) and its thermal decomposition fractions produced by ramped pyrolysis/oxi-

dation for a core retrieved from the >7.5 km-deep Japan Trench. High-resolution 14C mea-

surements, coupled with distinctive thermogram characteristics of OC, reveal hemipelagic

sedimentation interrupted by episodic deposition of pre-aged OC in the trench. Low δ13Cvalues and diverse 14C ages of thermal fractions imply that the latter material originates from

the adjacent margin, and the co-occurrence of pre-aged OC with intervals corresponding to

known earthquake events implies tectonically triggered, gravity-flow-driven supply. We show

that 14C ages of thermal fractions can yield valuable chronological constraints on sedimentary

sequences. Our findings shed new light on links between tectonically driven sedimentological

processes and marine carbon cycling, with implications for carbon dynamics in hadal

environments.

DOI: 10.1038/s41467-017-02504-1 OPEN

1 Geological Institute, ETH Zurich, 8092 Zurich, Switzerland. 2 National Ocean Science Accelerator Mass Spectrometry Facility, Woods Hole OceanographicInstitute, Woods Hole, MA 02543-1539, USA. 3 Institute of Geology, University of Innsbruck, 6020 Innsbruck, Austria. 4MARUM-Center for MarineEnvironmental Sciences University of Bremen, D-28359 Bremen, Germany. 5 Laboratory for Ion Beam Physics, Department of Physics, ETH Zurich, 8093Zurich, Switzerland. 6 Scottish Universities Environmental Research Centre, Glasgow G75 0QF, UK. 7Present address: Department of Earth and PlanetarySciences, Harvard University Cambridge, MA 02138, USA. Correspondence and requests for materials should be addressed toR.B. (email: rui.bao@erdw.ethz.ch) or to M.S. (email: Michael.Strasser@uibk.ac.at) or to T.I.E. (email: timothy.eglinton@erdw.ethz.ch)

NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications 1

1234

5678

90

The deep sea represents one of the most extreme and least-well-studied environments on Earth. Ecological, biogeo-chemical, and sedimentological processes of the hadal zone

(>6 km water depth)1 are particularly poorly constrained, yet offundamental importance for understanding the significance ofthese spatially expansive environments on a global scale2–6.Typical hadal environments are deep ocean trenches formed bythe downward bending of the oceanic lithosphere at subductionzones along active convergent plate boundary systems, such asaround the Pacific rim of fire7. Active convergent margins release>90% of the stress accumulated by global plate tectonics in oftendevastating earthquakes associated with the process of subduc-tion8. Recent earthquakes along subduction zones have providedan opportunity to understand dynamical earthquake-triggeredsediment remobilization processes in these hadal zones, and todetermine spatial-temporal characteristics of such events depos-its9–11. Key challenges for understanding long-term magnitude-frequency relationships of great subduction zone earthquakes andfor assessing their hazard potential are to reliably identify anddate earthquake-related event deposits in the geological record, toconstrain the provenance and frequency of these events deposits,and to link them to relevant earthquake parameters of the causalseismic events. However, one of the most critical challenges anddistinctive feature of the hadal zone environments such as deep-sea trenches, is that the underlying sediments are deposited belowthe calcite compensation depth (CCD), resulting in an absence ofdateable inorganic matter (i.e., carbonate biominerals), therebyconfounding traditional radiocarbon (14C) dating methods. 14Cmeasurements on the associated organic matter (OM), which isnot subject to the influence of dissolution of carbonate materials,can provide an alternative approach in these settings. Thisapproach has been previously applied to oceanic environmentsthat are depauperate in carbonate biominerals such as the Ant-arctic Ocean12–15. However, dilution of OM derived from marineproductivity with uncertain and variable proportions of pre-agedor petrogenic OM can undermine the validity of this approach.

While there is a growing understanding of sediment supply tothe abyssal ocean trenches5,9,16, our knowledge of the source andcomposition of associated OM deposited in hadal sedimentsremains very limited. Only a very small fraction of the OMproduced in surface waters escapes remineralization in the watercolumn, settles to the abyssal ocean floor, and is eventually bur-ied17–19. In deep ocean trenches associated with subductionzones, OM derived from pelagic sedimentation may be aug-mented by sediment supply from the adjacent margin. The latter,previously deposited and stored in upslope settings, may besupplied via background sedimentation processes (e.g., bottomand intermediate-depth nepheloid layer transport (BNL andINL)) or via more episodic gravity flows20. Thus, OM accumu-lating in trench sediments may contain a mixture of organiccarbon (OC) derived from autochthonous contemporary marineproductivity, as well laterally transported OC comprised of con-tinentally derived OC and reworked marine OC. The latterallochthonous inputs may be pre-aged as a consequence of theirpre-depositional histories (storage in intermediate reservoirs onthe continent and/or the margin), leading to 14C age offsets withautochthonous OM. Consequently, hadal zone sediments maycontain OM that varies in age and reactivity (bioavailability)2 as afunction of its provenance and mode of supply.

Recently, several new chronological approaches have beendeveloped for marine sedimentary OM that obviate complicationsfrom mixed OM sources. For example, compound-specificradiocarbon dating21,22 has been successfully applied to Antarc-tic Ocean sediments14. However, the application of the method isoften limited due to low concentrations of target compounds.Another approach uses a so-called ramped pyrolysis/oxidation

(RPO) method in combination with carbon isotopic analysis ofthe CO2 evolved from thermal decomposition of the sedimentaryOM15,23. This approach can yield both radiocarbon and stablecarbon isotopic information on sedimentary OM componentsseparated according to their thermochemical stability, revealingthe spectrum of 14C ages, and hence the heterogeneity of OCwithin a sample. When coupled with new methodologies thatallow for high-throughput bulk sediment OC 14C determina-tion24, it may yield more robust constraints on sedimentchronologies.

Sediment remobilization induced by the Tohoku-oki earth-quake (moment magnitude >9) and associated tsunami thatstruck NE Japan on 11 March 2011 triggered dense nepheloidlayers in the >7 km-deep Japan Trench9 and resulted in char-acteristic event deposits in underlying sediments10,11,25. TheJapan Trench is an oceanic trench formed by the subduction ofthe oceanic Pacific Plate below the Okhotsk Plate26. This plateboundary system is an active seismogenic zone that hosted themost recent tsunamigenic mega-earthquake27, triggering wide-spread remobilization and subsequent re-deposition of sedimentand associated OM in confined terminal basins with water depths>7 km, far below the CCD9,10,25.

Building on emerging knowledge of earthquake-triggeredsediment remobilization processes gained from this recentearthquake, and the strong temporal constraints on prior similarevents25, the objectives of this study were to assess the role ofsediment mobilization and lateral transport processes on thecharacteristics and 14C age of OM in hadal sediments, examine14C age-depth profiles of hadal sediments in the context of epi-sodic, event-driven sedimentation, and assess whether reliableOM-based 14C chronologies and age models can be establishedfor carbonate biomineral-poor hadal zone sediments. We explorethe utility of the RPO method for determining the radiocarbonages and carbon isotopic compositions of OM in Japan Trenchsediments, and for assessment of remobilization processes alongthis tectonically active subduction margin. Thermograms andcorresponding RPO 14C age spectra from sediments sampled atselected depth intervals are presented in the context of a high-resolution bulk OC 14C record. The origin and fate of OC, as wellas identification of tectonic events that episodically deliver pre-aged OM to Japan Trench sediments are discussed. This con-stitutes the first application of the RPO method and high-resolution OC 14C age profiling to such hadal zone sediments.Our observations reveal translocation and burial of significantquantities of pre-aged OC in the hadal environment, sheddingnew light on the nature and dynamics of carbon supply to hadalzone.

ResultsHigh-resolution bulk OC 14C profile. A high-resolution bulkOC 14C depth profile for core GeoB 16431-1 revealed large var-iations in age, from 1651± 81 to 9769± 196 14C yr BP (Figs. 1 and2, Supplementary Data 1). Marked 14C age excursions are evidentfor two depth intervals (~ 260–425 and ~ 500–616 cm) thatcoincide with turbidite sequences triggered by the AD 1454 and AD

869 earthquakes, respectively25,28. Strong linear fits exist throughthe three intervening depth intervals defined by Ikehara et al.25,(Fig. 2; red lines, overall r2> 0.8).

In-depth geochemical analysis. Five samples representing dif-ferent sedimentation phases (A–E; Fig. 1) were selected for fur-ther characterization (Supplementary Table 1). Total OC (TOC)values range from 0.6 to 1.7%, (ave., 1.3± 0.4%; n = 5), withhighest values for sample A and lowest values for samples B andD. Bulk OC δ13C values range from –30.9‰ to –25.1‰ (ave.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1

2 NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications

–27.7± 2.3‰; n = 5), with the lowest values for samples B and D.The δ13C profile shows relatively consistent values except withinthe earthquake sequences (Supplementary Fig. 1). Measured bulkOC 14C ages (Supplementary Table 1) range from 1850± 107 yrBP (sample A) to 9425± 228 yr BP (sample D). The 14C age ofsedimentary OC from the shallowest core depth (sample A, 5–7cm, 1850 14C yr BP), corresponding to the period of BNLdeposition associated with the 2011 Tohoku-oki earthquake, ismarkedly older than both that of sinking particulate matterintercepted at 8681 m in the hadal zone of the Japan Trench (~250 14C yr BP29), and ~ 1900 14C yr older than sediments

immediately below this earthquake sequence. Additionally, the14C age of sample B (352–355 cm; 5218± 146 yr BP), corre-sponding to a layer inferred to have been deposited in relation toa major tectonic event25, is older than sample C (448–451 cm;3139± 120 yr BP) deposited earlier. Similarly, the 14C age ofsample D (544–548 cm; 9425± 228 yr BP), the third intervalhypothesized to reflect a past earthquake event, is older than theunderlying sample E (615–618 cm; 3228± 122 14C yr BP).

The thermograms from RPO analysis of samples A–E in eachcase show a bimodal distribution with temperatures of maximumCO2 generation (Tmax) of 340± 10 °C (peak 1) and 450± 8 °C

1

2

3

4

5

6

7

Dep

th (

m)

vf f m c

San

dS

ilt

Cor

elit

holo

gy

Oxidized layer

Massivev v

vvv

v

140° 144°

36°

38°

40°

42°

00

NE

Jap

an

120

200

300

400

500

600

700

800

900

1

2

3

4

5

0

Ramped temperature (°C)

120

200

300

400

500

600

700

800

900

Temperature (°C)

1

2

3

4

5

0

AD 2011 Tohoku-oki earthquake

AD 1454 Earthquake

AD 915 Volcano eruption

AD 869 Jogan earthquake

Event

0

1

2

3

4

5

1

2

3

4

5

0

0

1

2

3

4

5

GeoB16431–1

T1

T2 T3

T4T5

T1

T2 T3

T4T5

T1

T2 T3

T4T5

T1T2

T3

T4 T5

T1

T2

T3

T4T5

Ash patchv v

vvv

v

v vv

vv

v

Volcanic ash

Sand layer

Coarse silt layer

Ripple (cross-) lamination

Ash patch

Fining upward

Sample A 5–7 cm

Sample C 448–451 cm

Sample E 615–618 cm

Peak 1Peak 2

Peak 1Peak 2

Peak 1 Peak 2

Peak 1

Peak 2

Peak 1 Peak 2

Nor

mal

ized

CO

2 (p

pm ×

10–4

)N

orm

aliz

ed C

O2

(pp

m ×

10–4

)

Homogenouse,coccolith-bearingdiatomaceous mud

Bioturbated diatomaceous mud

Homogeneousdiatomaceous mud

200 km

ab

c

d

e

g

Nor

mal

ized

CO

2 (p

pm ×

10–4

)

120

200

300

400

500

600

700

800

900

Temperature (°C)

Nor

mal

ized

CO

2 (

ppm

×10

–4)

Nor

mal

ized

CO

2 (

ppm

×10

–4)

GeoB 16431-1(38° 00.177′ N, 1143° 59.981′ E, 7542 m)

120

200

300

400

500

600

700

800

900

Temperature (°C)

120

200

300

400

500

600

700

800

900

Temperature (°C)

f

Sample B 352–355 cm

Sample D 544–548 cm

–4000

–6000

–6000

–2000

Fig. 1 Location and lithology of core GeoB 16431-1 in the Japan Trench, and thermograms of RPO of five sub-samples. a Map with location of core GeoB16431-1 (7542m water depth) in the Japan Trench, where the sediment sequence records historical earthquakes and volcanic eruptions. b Lithology of coreGeoB 16431-1 and depths of samples (A–E) selected for in-depth study are presented; samples A, B, and D are from inferred earthquake deposits,characterized by graded fine-sand layers fining-upward into homogenous mud, that are sedimentologically distinct from the bioturbated muds that reflectbackground sedimentation (samples C and E)46. The colored boxes show tectonic events25. These event deposits correlate with prior earthquakes (AD2011, AD 1454, and AD 869) and volcanic eruptions (AD 915) documented in Japanese historical records25. c, d, e, f, g Thermograms from RPO analysis ofsamples A–E are shown; Y-axis is mass-normalized CO2 concentration, X-axis indicates temperature gradient

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications 3

(peak 2), respectively (Fig. 1). While this suggests similarities inoverall thermochemical stability, and hence bulk OM character-istics, there are variations in relative peak height indicatingdifferences in proportions of different OM constituents. Insamples A, C, and E, the heights of first (lower temperature)peaks are proportionally higher than those of samples B and D(Fig. 3). Stable carbon isotopic values for each thermal fractionwere within the range expected for predominantly terrestrial OM(Supplementary Table 1 and Supplementary Fig. 3). In general,the δ13C values decrease from low to high temperatures, rangingfrom –36.5‰ to –21.9‰ (ave., –26.6± 3.7‰, n = 25). The 14C ageof thermal fractions also varies considerably, ranging from 1410± 15 to 13553± 193 yr BP. In general, lower temperature fractionsexhibit younger 14C ages, with older 14C ages for highertemperature fractions. Consistently older 14C ages are evidentfor sample D, with the highest temperature fraction for thissample (T5) yielding a radiocarbon age of 13553± 193 yr BP. Incontrast, the 14C ages of all thermal fractions from sample A(ranging from 1410± 15 to 4588± 93 yr BP) are youngercompared to corresponding fractions from other samples.

DiscussionRadiocarbon measurements indicate that pre-aged OC is depos-ited in the Japan Trench, particularly during tectonically triggeredsedimentation. Samples B and D have markedly older 14C agescompared to other intervals (Supplementary Table 1), suggestingentrainment of old OC or enhanced proportions of recalcitrantand old OC due to preferential degradation of labile OC. The firstexplanation is inconsistent with observations from RPO analysis(Fig. 1), which showed that all five thermogram patterns weresimilar, suggesting common source(s) of OM to sediments in theJapan Trench. In contrast, the varying proportions of low- andhigh-temperature peaks are attributed to differing extents ofdegradation of OM components within each sample. Notably,while the present data set remains limited, a strong linear

relationship is apparent between bulk OC 14C content (Fm) andthe height ratio of peak 1 to peak 2 (Fig. 3), suggesting that bulk14C content is largely controlled by the relative proportion oforganic components. Such changes in the relative proportion oforganic components (peaks) can also be interpreted in the contextof selective degradation/preservation of OM. We infer that therelatively old 14C ages of samples B and D result from longerresidence times within diagenetically active reservoirs (watercolumn and surface sediments) and hence greater degradationprior to burial. Stable carbon isotopic characteristics (Supple-mentary Table 1), specifically the 13C depletion of bulk OM(δ13C: –25.1‰ to –30.9‰) compared with the typical δ13C valuesof sinking particulate OM (~−23.0‰ and ~−23.5‰ at 4789 and8789 m water depth, respectively29), suggests that terrestrial OM(or possibly refractory aliphatic macromolecular material30) is thedominant source of sedimentary OC in the Japan Trench. Inparticular, δ13C values of samples B and D (–28.7‰ and –30.9‰,respectively, Supplementary Table 1), corresponding to the AD

1454 and 869 tectonic events, respectively, are ~3‰ lower thanthose of other samples. Associated OM may therefore be subjectto preferential loss of labile, 13C-enriched marine OM duringlateral transport and subsequent sedimentation in the Trench.Anomalina spp., a contemporary shallow-water benthic for-aminiferal species, was found in sample B, suggesting sedimenttransfer and burial following a tectonic event that triggereddelivery of continentally derived sediments. This is also consistentwith the inferred sedimentation mode25 whereby earthquake-triggered gravity flows rapidly remobilize and translocate pre-viously deposited sediment and its associated OM from thecontinental margin to the deep Japan Trench. Gravity flowstriggered by tectonic events such as earthquakes (and resultanttsunamis and landslides) lead to relatively high sedimentationrates in these abyssal settings (refs 10,11,25, and referencestherein). The interseismic hemipelagic deposits that form withhigh sedimentation rates from 0.8 to >3.0 m kyr−1 effectivelycover earthquake-induced turbidites and volcanic ash layers andpreserve the deposits as a geological record of large tectonic andvolcanic events25. Ikehara et al.25 documented thick (~1.5 m),fining-upward turbidite deposits and volcanic ash layers inter-bedded within bioturbated diatomaceous mud in sediment coresfrom this region. The terminal basins along the trench flooraccommodate the episodic deposition of fine-grained turbidites asgraded fine-sand layers fining-upward into homogenous diato-mateous mud. In general, coarser-grained sediments sink more

102 103

OC fluxes (tons km–2 yr–1) -NM

0

1

2

3

4

5

6

7

104 1052 4 6 8 10 120

1

2

3

4

5

6

7

14C age (×103 yr BP)

AD 2011 Tohoku-oki earthquake

AD 1454 Earthquake

AD 869 JoganEarthquake

Dep

th (

m)

D: 544–548 cm

B: 352–355 cm

E: 615–618 cm

C: 448–451 cm

A: 5–7 cm

Other samples

Fig. 2 Lithology of core GeoB 16431-1. A high-resolution bulk OC 14C ageprofile and corresponding OC fluxes (note log scale)

Height ratio of peak1 to peak 2

Bul

k O

C F

m

r 2 = 0.95

D: 544–548 cm

B: 352–355 cm

E: 615–618 cm

C: 448–451 cm

A: 5–7 cm

0.6 0.7 0.8 0.9 1 1.1 1.20.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

y = 0.98×–0.45

P < 0.001

Fig. 3 Relationship between fraction modern of bulk OC and height ratio ofpeak 1 to peak 2 in the RPO thermogram

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1

4 NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications

rapidly, whereas the finer-grained sediments settle more slowly,resulting in upward-fining of sediments within the event deposit(Fig. 1, lithology). Indeed, some fractions of the sedimentsintroduced into suspension may reside for protracted periods oftime within the BNL following an episodic event31,32. Samples Band D are from the upper (finer-grained) zone, implying later-stage deposition (Fig. 1), and protracted entrainment withinBNLs (or INLs).

The relatively invariant temperatures for maxima of the firstand second RPO peaks (Tmax1 and Tmax2; 340± 10 and 450± 8 °C, respectively) suggest that there is little difference in thermo-stability, and by inference the characteristics of the OM, due topost-depositional diagenetic processes in the Japan Trench sedi-ment core. The nature of OC (e.g., lability vs. recalcitrance) can beindirectly inferred based on thermal stability33. The samplesinvestigated here have peaks presumably corresponding to the“labile” and “recalcitrant” OC categories. The relatively old 14Cages of corresponding thermal fractions T2 (320–391 °C) and T3

(391–476 °C) (ranging from 2050± 85 to 12164± 145 yr BP)suggest burial of old yet both relatively labile and recalcitrant OMin the Japan Trench. Preferential degradation of young, labile OMmay lead to an increase bulk 14C ages. Figure 4 shows the 14C agespectrum of thermal fractions among the five samples investi-gated. The 14C age difference (14C heterogeneity) between themost labile (lowest temperature, T1) and the most refractory(highest temperature, T5) fraction exceeds 4000 yr in samples Band D. In contrast, samples C and E (inferred background sedi-mentation, deposited around AD 915 and 869, respectively)exhibit smaller and more consistent age offsets (2714± 118 and2820± 122 yr, respectively; Fig. 4). Preferential degradation alsoenhances 14C age heterogeneity (i.e., samples B and D). Thedegree of 14C age heterogeneity may depend on the pre-depositional history and/or extent of in situ microbial degrada-tion of younger OC.

The high-resolution bulk OC 14C profile provides furtherinsights into temporal variability in sedimentary OM accumula-tion in the Japan Trench (Fig. 2). Two depth intervals that arecharacterized by markedly older 14C ages, ~ 260–425 and ~500–616 cm, correspond well to sedimentological layers attrib-uted to the AD 1454 and AD 869 earthquakes, respectively25,implying enhanced (re)burial of OC in the trench associated withthese events. This is also consistent with RPO results that indi-cates greater accumulation of old OC in the Japan Trench

triggered by tectonic events. These older ages could be a con-sequence of protracted storage in intermediate reservoirs on land(e.g., soils) and/or on the continental margin. For example, OMin soils and exported to the ocean from rivers can exhibitapparent residence times >1000 yr34,35. OM can reside withincontinental margin surficial sediments for millennia36. Thesediment drape over continental slopes can be destabilized bytectonic activity11 and mobilized sediments may remainentrained in suspension-deposition cycles and subject to wide-spread dispersal via nepheloid layer transport for several hundredto several thousand years36–38 prior to eventual re-deposition.These processes may be accompanied by OM degradation and14C aging. Irrespective of the specific process(es) at play, theseresults indicate that remobilization of earthquake-triggered sedi-ments results in enhanced burial of pre-aged OC in the JapanTrench (Fig. 2, OC fluxes).

The bulk 14C age of the uppermost sediment layers from thegravity core (ave. 1811 yr BP, n = 3, top 20 cm) is ~ 2000 yr olderthan the date of sediment emplacement following the AD 2011Tohoku-oki earthquake. The markedly older 14C ages of OC insediments deposited coincident with prior tectonic events alsoimplies significant and time-varying inputs of margin-derivedpre-aged OC (Fig. 2). These variable contributions of allochtho-nous OC confound to develop 14C-based chronologies for sedi-mentary sequences deposited in the hadal zone, and precludedirect age assignments for specific event layers. The influence ofaged OC in Japan Trench sediments is not restricted to event-related deposition (Fig. 2). For instance, the documented tephraage is AD 915, whereas the bulk OC 14C age just below the tephralayer is 3139± 120 yr BP, clearly exceeding the estimated localmarine reservoir age of 830 yr39. Background sedimentation (i.e.,excluding the event layers) is characterized by quasi-linear rela-tionships between bulk OC 14C age vs. sediment depth (Fig. 2, redlines), suggesting that in the absence of major tectonic events,similar sedimentological and diagenetic conditions prevail overthe duration of the record, and implying that OM originates fromcommon or similar sources. However, development of sedimentchronologies for Japan Trench sediments based on bulk OC 14Cage is prone to uncertainty due to variable and uncertain pro-portions of aged OC.

Although bulk OC 14C ages of Japan Trench sediments areinsufficiently constrained to derive chronologies, it may be pos-sible to use 14C ages of specific (thermal) fractions from RPO of

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Radiocarbon age (×103 yr BP)

Depth (m)

6

5

4

3

2

1

AD 2011

AD 1454AD 915

AD 869

1096 yr

46 yr

EventsAD 2011 Tohoku-oki earthquake

AD 1454 earthquake

AD 869 Jogan earthquake

AD 915 volcano eruption

T1 (170–320 °C)

T2 (320–391 °C)

T3 (391–476 °C)

T4 (476–570 °C)

T5 (570–915 °C)

A

B

C

D

E

Fig. 4 Chronology and radiocarbon characteristics of RPO thermal fractions of OM in sediment core GeoB 16431-1 for samples A–E. The samples’ labels areshown adjacent to right-hand Y-axis and in Fig. 3. Adjacent to the right-hand Y-axis are the thermal decomposition temperature intervals. The gray shadingdenotes the earthquake events. Their names and extents are marked by arrows on the left-hand Y-axis. The X-axis (top) indicates the 14C age (yr BP) ofthermal decomposition fractions (the measurement error is smaller than the symbol)

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications 5

samples derived from different depth horizons to establish arelative chronology. The RPO results reveal that the lowesttemperature fractions (T1, labile fraction) are older than theknown 14C ages of the tectonic events. For example, the con-ventional 14C age of T1 from sample C (below the AD 915 volcanoeruption) is 2830± 20 yr BP. While it is not presently possible toaccurately determine the age and proportion of autochthonousand allochthonous OC in a sample, precluding use of conven-tional 14C ages of both bulk OC and T1 fraction as a absolutedating tools, 14C age differences between specific thermal frac-tions may provide an approximation of age differences betweensediment intervals, particularly for sediments deposited withinthe last ~ 3000 yrs, during which radiocarbon ages closely parallelcalendar ages (INTCAL1340). We note that the 14C age difference(120± 45 yr) between T1 fractions in samples C and E approachesthe time offset (46 yr) between the volcanic eruption in AD 915and Jogan earthquake in AD 869 (Fig. 4), particularly consideringsampling uncertainty (samples C and E lie below each eventsequence). Similarly, the age offset (126± 195 14C yr) between T5

fractions from these two samples approximates the known timeoffset between the two tectonic events. The consistency in 14Cdifferences (in Fig. 4, if connected, they would be “parallel” lines)between sedimentary layers of samples C and E for both theyoungest thermal fraction (T1) and oldest thermal fraction (T5),and their correspondence with known tectonic and volcanicevents, implies that these thermal fractions are not affected bypreferential OM degradation during subsequent burial, while thelack of observed consistency in 14C ages of T3 fractions is likelybecause this temperature fraction is sensitive to subtle variationsin proportions of different organic components that contribute tothis specific temperature interval (391–476 °C). Overall, 14C agedifferences between corresponding thermal fractions from dif-ferent sediment intervals are inferred to primarily reflect radio-active decay following sedimentation. Considering measurementerrors and additional uncertainties, we believe this approachholds promise for placing chronological constraints on JapanTrench sediment cores. Age models could be constructed pro-vided these 14C ages can be anchored to at least one sample ofknown calendar age (e.g., dated tephra layer).

It is interesting to note that the 14C age of OC in the upper-most sediment interval investigated, which represents depositionfollowing the recent Tohoku earthquake (AD 2011), is relativelyyoung compared to the other layers attributed to previousearthquakes (AD 1454 and 869). This could be explained if fine-grained sediments carrying older OM remain in suspension for along time prior to eventual deposition9,32. Alternatively, varia-tions in provenance of sediments delivered to the Japan Trench,which depends on the upslope location(s) of sediment failure andgravity flow, may be responsible for these different age char-acteristics. Finally, old OC ages could be masked by supply ofrecent OM that has inherited an elevated bomb radiocarbonsignal via photosynthetic carbon fixation in the surface ocean41.Further investigation of sedimentary processes following the mostrecent earthquake is warranted in order to be able to recognize,and to constrain the frequency and origin of past tectonic eventsin hadal sedimentary sequences.

This first application of RPO and 14C analysis of specificthermal decomposition fractions provides new chronologicalconstraints on past depositional events in the Japan Trench. Theapproach is particularly useful when coupled with high-resolutionbulk OC 14C measurements that document when backgroundhemipelagic sedimentation is interrupted by tectonically triggeredgravity flows. Our observations highlight the large-scale translo-cation and burial of terrigenous materials in the hadal zoneassociated with these events. This lateral carbon pump hasimplications for nature and dynamics of carbon supply to abyssal

ocean subduction zone sediments, and for associated benthiccommunities2. More broadly, this approach holds promise fordevelopment of chronologies for hadal and other sediments thatlack microfossils for conventional radiocarbon dating and isotopestratigraphy.

MethodsSampling strategy and preparation. We analyzed gravity core GeoB 16431-1 (9.4m recovery), which was retrieved from a terminal Japan Trench basin during R/VSONNE cruise SO219A (ref. 28; sampling site 38° 00.177′ N, 143° 59.981′ E; 7542 mwater depth; Fig. 1, map). The core was split, described, and sectioned on board.The sediment core sections were then shipped and stored at 4 °C in the corerepository of MARUM-Center for Marine Environmental Sciences at the Uni-versity of Bremen, and subsequently subsampled and stored at –20 °C at ETHZurich until further processing.

A suite of 82 sediment samples (representing sampling interval of ~ 7.5 cm;Supplementary Data 1) was selected for construction of a high-resolution profilesof bulk OC content and 14C age. Samples were freeze-dried in pre-combusted vials,and aliquots were weighed into Ag capsules for fumigation with concentrated HCl(37% Trace-Metal purity, 60 °C, 72 h) to remove inorganic carbon. The acidifiedsamples were subsequently neutralized with NaOH under the same conditions (72h). Sample preparation was performed in the Biogeoscience Group Laboratories atETH Zurich (Fig. 2; Supplementary Data 1).

Five sub-samples from specific depth intervals were selected for in-depthanalysis (Supplementary Table 1). The sampling strategy was guided by theinterpreted event stratigraphy of this core25, inferred sediment remobilizationevents (i.e., sample A (5–7 cm) deposited following the 2011 Tohoku earthquake;samples B (352–355 cm) and D (544–548 cm) from older event deposits linked tohistorical earthquakes25); and of intervals of quasi-continuous backgroundsedimentation in the vicinity of marker beds with known depositional age (sampleC (448–451 cm) just above the To-a volcanic ash layer of AD 915, and sample E(615–618 cm) just below the event layer correlated to the Jogan Earthquake of AD

86925; Fig. 1). Samples for RPO measurements were prepared in the same manneras above, but in the National Ocean Science Accelerator Mass Spectrometry facility(NOSAMS) at Woods Hole Oceanographic Institution.

Bulk organic geochemical properties. TOC content of the five selected sedimentsamples was determined using established protocols at NOSAMS42. Freeze-driedsamples (~ 3 g dry weight) were weighed into pre-combusted glass Petri dishes (5cm diameter). A beaker filled with ~ 30 mL HCl (37%, Trace-Metal purity) wasplaced at the bottom of a 250 mm inner diameter glass desiccator; the samples wereplaced on a ceramic tray above the acid beaker. The desiccator was evacuated andsamples were treated at 60 °C for 72 h. After fumigation, excess acid was neu-tralized through replacing the acid beaker with ~ 20 g NaOH pellets in a pre-combusted Petri dish. The desiccator containing the acidified samples with NaOHpellets was again evacuated and placed in an oven (60 °C, 72 h). Approximately 7%of the purified CO2 gas was split on a vacuum line and used for stable isotopedetermination via isotope ratio mass spectrometry (IRMS), with the remaindergraphitized using standard methods at NOSAMS43. Corresponding stable carbonisotopic (δ13C) values were determined to a precision of better than ±0.1‰, andare reported relative to Peedee Belemnite. Three samples listed in SupplementaryTable 1 were measured for 14C at NOSAMS, while the other two were analyzed for14C using a MICADAS accelerator mass spectrometer (AMS) system at ETHZurich and calibrated against standard Oxalic Acid II (NIST SRM 4990C) and in-house radiocarbon blank CO2 or anthracite coal. Samples for 14C analysis at ETHZurich were measured directly as CO2 gas. 14C precision for gas was better than±10‰ on a modern standard. The high-resolution 14C ages of 82 bulk samples’analysis and 5 bulk samples from specific depth intervals were measured using acoupled elemental analyzer (EA)/IRMS/AMS online system at ETH Zurich24.Radiocarbon data are reported as fraction modern (Fm) and radiocarbon age44.

RPO measurements. Ramped temperature pyrolysis/oxidation was performed onthe five bulk sediment samples selected for in-depth analysis (SupplementaryTable 1; Fig. 4). As in previous work15,23, acidified samples (~ 100–200 mg) wereloaded into a quartz reactor and subjected to a constant rate of temperatureincrease (5 °Cmin−1) until 915 °C. Evolved components were simultaneously oxi-dized and the resulting gases (e.g., CO2, SO2) were purified by passage through achemical reactor under isothermal conditions45. CO2 concentrations and theresulting thermograms were obtained using a flow-through infrared CO2 analyzer(Sable Systems International Inc., CA-10a). Evolved thermal components wereintegrated over five temperature intervals Tn (T1: 170–320 °C; T2: 320–391 °C; T3:391–486 °C; T4: 486–570 °C; and T5: 570–915 °C; Fig. 1c–g). A leak check wasconducted at the beginning and every hour during the experiment. CO2 samplescorresponding to individual thermal fractions were sealed in pre-combusted glasstubes with copper oxide and silver balls for combustion to purify gases prior to 14Cmeasurement.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1

6 NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications

OC fluxes. OC fluxes were calculated using measurements of sample density, TOCcontent, and sedimentation rate (OC flux = TOC×density×sedimentation rate).Sample density (bulk density, g cm−3) was measured by gamma-ray attenuationdensity analyses from multi-sensor-core logging at MARUM (SupplementaryData 1). TOC was measured using the EA/IRMS/AMS system at ETH Zurich (datashown in Supplementary Data 1). Sedimentation rate was calculated based on theratio of sediment depth spanning time intervals constrained by known events.Sedimentation rate for the 2011-earthquake turbidite sequence was calculated bythe ratio of depth of sequence (~ 30 cm) to 1 yr (time interval between 2011 and thecollection year, ~ 30 cm yr–1). Sedimentation rates for the other two earthquaketurbidite sequences were estimated by multiplying ~ 30 cm yr−1 and each sequencedepth (assuming that every earthquake sequence exhibits rapid sedimentationduring 1 yr). Due to compaction during the sedimentation, our approach may yieldunderestimates of the OC fluxes associated with the AD 1454 and AD 869earthquakes.

Chronological considerations. In the section of this paper addressing the devel-opment of Japan Trench sediment chronologies, we focus on 14C offsets betweenthermal fractions (e.g., samples C and E) in order to derive a relative, rather thanabsolute chronology, and highlight its potential relevance to other hadal zonesedimentary sequences lying below the CCD that lack microfossils for conventionalradiocarbon dating and isotope stratigraphy. Although the differences in isotopiccompositions between bulk OC and RPO data are significant, they are roughlyconstant (small, Fm values ~ 0.06–0.08, ±0.02; Supplementary Fig. 2). As such, theisotopic discrepancy does not impact the chronological aspect of the paper andspecifically the validity of chronologies developed based on RPO. Considering themass contribution of target thermal fractions to each sample and the associated 14Cmeasurement errors (ave. ±~100 14C yr, based on ETH measurement), the offset(relative value) between 14C ages of thermal fractions is relatively small (±~200 14Cyr), and does not undermine our statement in the manuscript. In the manuscript,we acknowledge this time difference (i.e., relative offset: 120± 45 yr, time offset: 46yr). We emphasize that the novelty of our approach is to first identify whichsamples (background sedimentation) are appropriate for further chronologicalanalysis through bulk OC 14C measurement, and use information derived fromRPO to constrain 14C age offsets (of T1 fractions) between sediment layers. Thisyields relative ages that can be anchored based on one or more known events(calendar age). Furthermore, while note that the presented 14C ages are notcalendar ages (documented event time), during the late Holocene (which con-stitutes the time interval of primary interest in this study), measured 14C agesclosely parallel calendar ages40. Thus, accuracy of the relative chronology isdependent on measurement precision and offsets between calendar and 14C ages,we consider this constitutes a unique approach for documenting the timing andfrequency of event deposits pre-dating historical records in the hadal zone.

Data availability. Data sets generated during and/or analyzed during the currentstudy can be found in Supplementary Table 1 and Supplementary Data 1, and arealso available from the corresponding author on reasonable request.

Received: 23 December 2016 Accepted: 5 December 2017

References1. Wolff, T. The hadal community, an introduction. Deep Sea Res. 6, 95–124

(1960).2. Glud, R. N. et al. High rates of benthic microbial activity at 10900 meters depth:

Results from the Challenger Deep (Mariana Trench). Nat. Geosci. 6, 284–288(2013).

3. Jamieson, A. J., Fujii, T., Mayor, D. J., Solan, M. & Priede, I. G. Hadal trenches:the ecology of the deepest places on Earth. Trends Ecol. Evol. 25, 190–197(2010).

4. Jamieson A. J. The Hadal Zone: Life in the Deepest Oceans. 1–382 (CambridgeUniversity Press, UK, 2015).

5. Turnewitsch, R. et al. Recent sediment dynamics in hadal trenches: evidence forthe influence of higher-frequency (tidal, near-inertial) fluid dynamics. Deep SeaRes. I 90, 125–138 (2014).

6. Bartlett, D. H. Microbial life in the trenches. Mar. Technol. Soc. J. 43, 128–131(2009).

7. Clift, P. & Vannucchi, P. Controls on tectonic accretion versus erosion insubduction zones. Rev. Geophys. 42, 1–31 (2004).

8. Ruff, L. & Kanamori, H. Seismicity and the subduction process. Phys. EarthPlanet. Inter. 23, 240–252 (1980).

9. Oguri, K. et al. Hadal disturbance in the Japan Trench induced by the 2011Tohoku–Oki Earthquake. Sci. Rep. https://doi.org/10.1038/srep01915 (2013).

10. Strasser, M. et al. A slump in the trench: tracking the impact of the 2011Tohoku-Oki earthquake. Geology 41, 935–938 (2013).

11. McHugh, C. M. et al. Remobilization of surficial slope sediment triggered by theA.D. 2011Mw9 Tohoku-Oki earthquake and tsunami along the Japan Trench.Geology 44, 391–394 (2016).

12. Harris, P., O’brien, P., Sedwick, P. & Truswell, E. Late Quaternary history ofsedimentation on the Mac. Robertson Shelf, East Antarctica: problems with 14C-dating of marine sediment cores. Pap. Proc. R. Soc. Tasman. 130, 47–53 (1996).

13. Licht, K. J., Jennings, A. E., Andrews, J. T. & Williams, K. M. Chronology of lateWisconsin ice retreat from the western Ross Sea, Antarctica. Geology 24,223–226 (1996).

14. Ohkouchi, N., Eglinton, T. I. & Hayes, J. M. Radiocarbon dating of individualfatty acids as a tool for refining Antarctic margin sediment chronologies.Radiocarbon 45, 17–24 (2003).

15. Rosenheim, B. E. et al. Antarctic sediment chronology by programmed‐temperature pyrolysis: Methodology and data treatment. Geochem. Geophys.Geosyst. 9, Q04005 (2008).

16. Palanques, A. et al. Evidence of sediment gravity flows induced by trawling inthe Palamós (Fonera) submarine canyon (northwestern Mediterranean). DeepSea Res. I 53, 201–214 (2006).

17. Emerson, S. in Geophysical Monograph Series Vol. 32 (eds Sundquist, E. T. &Broecker, W. S.) 78–87 (AGU, Washington, D. C., 1985).

18. Hedges, J. I. & Keil, R. G. Sedimentary organic matter preservation: anassessment and speculative synthesis. Mar. Chem. 49, 81–115 (1995).

19. Honjo, S. et al. Understanding the role of the Biological Pump in the GlobalCarbon Cycle: an imperative for Ocean Science. Oceanography 27, 10–16 (2014).

20. Nittrouer, C. A. & Wright, L. D. Transport of particles across continentalshelves. Rev. Geophys. 32, 85–113 (1994).

21. Eglinton, T. I., Aluwihare, L. I., Bauer, J. E., Druffel, E. R. M. & McNichol, A. P.Gas chromatographic isolation of individual compounds from complexmatrices for radiocarbon dating. Anal. Chem. 68, 904–912 (1996).

22. Eglinton, T. I. et al. Variability in radiocarbon ages of individual organiccompounds from marine sediments. Science 277, 796–799 (1997).

23. Subt, C., Fangman, K. A., Wellner, J. S. & Rosenheim, B. S. Sedimentchronology in Antarctic deglacial sediments: reconciling organic carbon 14Cages to carbonate 14C ages using Ramped PyrOx. Holocene 26, 265–273 (2015).

24. McIntyre, C. P. et al. Online 13C and 14C gas measurements byEA–IRMS–AMS at ETH Zürich. Radiocarbon 58, 1–7 (2016).

25. Ikehara, K. et al. Documenting large earthquakes similar to the 2011 Tohoku-oki earthquake from sediments deposited in the Japan Trench over the past1500 years. Earth Planet. Sci. Lett. 445, 48–56 (2016).

26. Tsuru, T. et al. Along‐arc structural variation of the plate boundary at the JapanTrench margin: implication of interplate coupling. J. Geophys. Res. 107, ESE 11-1–ESE 11-15 (2002).

27. Suzuki, W., Aoi, S., Sekiguchi, H. & Kunugi, T. Rupture process of the 2011Tohoku-Oki mega-thrust earthquake (M9.0) inverted from strong-motion data.Geophys. Res. Lett. 38, L00G16 (2011).

28. Wefer, G. et al. Report and preliminary results of RV SONNE Cruise SO219A,Tohoku-Oki Earthquake-Japan Trench, Yokohama-Yokohama. MARUM-Zentrum für Marine Umweltwissenschaften, Fachbereich Geowissenschaften,Universität Bremen, 1–83, (Bremen, 2014).

29. Nakatsuka, T., Hosokawa, A., Handa, N., Matsumot, F. & Masuzawa, T. inDynamics and Characterization of Marine Organic Matter (eds Handa, N.,et al.) 169–186 (Springer, 1997).

30. Hwang, J. & Druffel, E. R. W. Lipid-like material as the source of theuncharacterized organic carbon in the ocean? Science 299, 882–884 (2003).

31. Arai, K. et al. Tsunami-generated turbidity current of the 2011 Tohoku-Okiearthquake. Geology 41, 1195–1198 (2013).

32. Bao, R. et al. Widespread dispersal and aging of organic carbon in thecontinental marginal seas. Geology 44, 791–794 (2016).

33. Lopez-Capell, E., De la Rosa Arranz, J. M., González-Vilab, F. J., González-Perezb, J. A. & Manning, D. A. C. Elucidation of different forms of organiccarbon in marine sediments from the Atlantic coast of Spain usingthermalanalysis coupled to isotope ratio and quadrupolemass spectrometry.Org. Geochem. 37, 1983–1994 (2006).

34. Torn, M. S., Trumbore, S. E., Chadwick, O. A., Vitousek, P. M. & Hendricks, D.M. Mineral control of soil organic carbon storage and turnover. Nature 389,170–173 (1997).

35. Drenzek, N. J., Montluçon, D. B., Yunker, M. B., Macdonald, R. W. & Eglinton, T.I. Constraints on the origin of sedimentary organic carbon in the Beaufort Sea fromcoupled molecular 13C and 14C measurements. Mar. Chem. 103, 146–162 (2007).

36. Keil, R. G., Dickens, A. F., Arnarson, T., Nunn, B. L. & Devol, A. H. What is theoxygen exposure time of laterally transported organic matter along theWashington margin? Mar. Chem. 92, 157–165 (2004).

37. Inthorn, M., Wagner, T., Scheeder, G. & Zabel, M. Lateral transport controlsdistribution, quality, and burial of organic matter along continental slopes inhigh-productivity areas. Geology 34, 205–208 (2006).

38. Hwang, J., Druffel, E. R. & Eglinton, T. I. Widerspread influence of resuspendedsediments on oceanic particulate organic carbon: insights from radiocarbon andaluminum contens in sinking particles.Global Biogeochem. Cycles 24, GB003802 (2010).

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications 7

39. Ikehara, K., Ohkushi, K., Noda, A., Danhara, T. & Yamashita, T. A new localmarine reservoir correction for the last deglacial period in the Sanriku region,northern North Pacific, based on radicoarbon dates form the Towada-Hachinohe (To-H) tephra. Quternary Res. 52, 127–137 (2013).

40. Reimer, P. J. et al. IntCal 13 and Marine 13 radiocarbon age calibration curves0-50000 year cal BP. Radiocarbon 55, 1869–1887 (2013).

41. McNichol, A. P. & Aluwihare, L. I. The power of radiocarbon inbiogeochemical studies of the marine carbon cycle: insights from studies ofdissolved and particulate organic carbon (DOC and POC). Chem. Rev. 107,443–466 (2007).

42. McNichol, A. P., Osborne, E. A., Gagnon, A. R., Fry, B. & Jones, G. A. TIC,TOC, DIC, DOC, PIC, POC-Unique aspects in the preparation of oceanogra-phic samples for 14C-AMS. Nucl. Instr. Met. Phys. Sect. B. 92, 162–165 (1994).

43. Vogel, J. S., Southon, J. R. & Nelson, D. E. Catalyst and binder effects in the useof filamentous graphite for AMS. Nucl. Instr. Met. Phys. Sect. B. 29, 50–56(1987).

44. Stuiver, M. & Polach, H. A. Discussion; reporting of 14C data. Radiocarbon 19,355–363 (1977).

45. Froelich, P. Analysis of organic carbon in marine sediments. Limnol. Oceanogr.25, 564–572 (1980).

46. Ikehara, K. et al. SP456. 9 (Geological Society of London, 2017).

AcknowledgementsWe thank the captain and crew of R/V SONNE for onboard assistance during cruiseSO219A in 2012. This cruise is supported by Federal Ministry for Education andResearch, and Deutsche Forschungsgemeinschaft. This study is supported by Doc.Mobility Fellowship (P1EZP2_159064) (R.B.) from the Swiss National Science Founda-tion (SNSF). This work is also supported by SNF “CAPS-LOCK” project 200021_140850(T.I.E.), by SNSF grant (133481) (M.S.), and Austrian Science Foundation (P 29678-N28)(M.S.). We thank Dr Lukas Wacker for helpful comments. We thank support of theNOSAMS staff in the execution of this project. We also thank Ms. Chen in TongjiUniversity for assistance in identification of foraminifera. We greatly appreciate theassistance from members of the Laboratory for Ion Beam Physics in all aspects of theAMS measurements.

Author contributionsT.I.E., M.S. and R.B. conceived and designed the project. R.B. wrote the paper with inputfrom all co-authors and produced the figures. M.S. and G.W. collected the samples. R.B.,T.I.E. and M.S. interpreted results. R.B., N.H., and C.M. carried out carbon isotopeanalyses. R.B. and A.P.M. carried out the RPO analyses.

Additional informationSupplementary Information accompanies this paper at https://doi.org/10.1038/s41467-017-02504-1.

Competing interests: The authors declare no competing financial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,

adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directly fromthe copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2017

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-02504-1

8 NATURE COMMUNICATIONS | (2018) 9:121 |DOI: 10.1038/s41467-017-02504-1 |www.nature.com/naturecommunications